Klasifikasi Minat Masyarakat Dalam Berlangganan Netflix di Masa Pandemi Covid-19 dengan Algoritma Naïve Bayes

Studi Kasus: Kecamatan Karawang Timur

Abstract

The first time the Covid-19 (Coronavirus Disease 2019) outbreak appeared was at the end of 2019 and the first time a Covid-19 case was found in Indonesia was in March 2020. Due to its fast spread, the World Health Organization (WHO) finally declared this outbreak a pandemic global. In order to prevent the spread of this epidemic, various countries implemented policies such as implementing social restrictions, lockdowns, and various other policies. With the ongoing Covid-19 outbreak, this has had quite a big impact on various sectors of life which of course has hampered the Indonesian economy. Netflix is ​​a Video on Demand application that is quite popular and has been established since 1997. Therefore, this research was compiled with the aim of obtaining results of the classification of the people's interest in subscribing to Netflix during the Covid-19 pandemic and to find out how many people use it. interested and not interested in subscribing to Netflix by applying one of the data mining techniques, namely classification with the naïve Bayes algorithm. The data in this study were obtained from the results of filling out a questionnaire in the form of a google form which was distributed on several social media such as Instagram, Twitter and Telegram and was filled out by the people of East Karawang District. The total sample data used is 399 records obtained from the calculation of the slovin formula. Then the data will be processed using the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology which consists of six stages. The 399 records will be divided into two, namely training data and testing data. The process of dividing training data and testing data will be grouped into three ratios, including 90:10, 80:20, and 70:30 which will then be applied to RapidMiner tools to determine accuracy, precision, recall, and AUC values. The results obtained from applying data to the RapidMiner tools show that the 90:10 ratio has better results compared to other ratios, where the accuracy value is 92.50%, the precision value is 92.11%, the recall value is 100%, and AUC of 0.857.